Depression and its related functional impairment and behavioral disturbances (including depression-related sleep disruption) pose clinically significant problems in patients with Alzheimer's dementia (AD), degrading quality of life, further burdening both patients and their caregivers, and possibly increasing the likelihood of early placement in long-term care. There are limited data supporting the efficacy of acute antidepressant pharmacotherapy in Alzheimer's dementia, but no data to guide the long-term treatment of depression and related behavioral burdens, to prevent the recurrence of depressive episode, and to prolong the community tenure of Alzheimer patients. Therefore, we propose a randomized, double-blind, placebo-controlled, variable-discontinuation trial of antidepressant medication (paroxetine as first-line agent; nortriptyline as second-line agent in paroxetine non-responders) to answer two questions: l) is antidepressive medication superior to placebo in bringing about recovery from depression and enhancement of function? and 2) is antidepressive medication superior to placebo in preserving recovery from depression? in addition, we propose to track rates of functional and cognitive change in Alzheimer patients in relation to recovery from depression and treatment assignment, in order to derive more precise information concerning the risk/benefit ratio of antidepressant pharmacotherapy in these patients. We predict that therapy with antidepressant medication will be superior to placebo in l) bringing about initial symptomatic and functional recovery from depression in AD; and 2) preventing or delaying the return of depressive symptoms and related functional impairment. We also predict that recovery from depression will be associated with improved cognitive performance. A total of 120 patients meeting DSM-IV criteria for Alzheimer's dementia with depression will be recruited and randomly assigned to one of three treatment cells: l) placebo; 2) paroxetine (26 weeks followed by placebo for 26 weeks), and 3) paroxetine (52 weeks). Recovery from and recurrence of major depression will be the primary outcome variables, but other outcomes will be examined to characterize course of illness. We will use survival analysis to test for differences in rates of recovery/recurrence and time to recovery/recurrence of depression in AD patients randomly assigned to the three treatment arms. Repeated-measure analysis of covariance, using depression scores as a time-varying covariate, will test for interaction effects between treatment and time on measures of cognition and function, care-giver strain, and duration of community tenure.